Article - May 2002
[Home Page]

[an error occurred while processing this directive]
(Click Message to Learn More)

Optimizing All-Variable Speed Systems with Demand Based Control
To ensure that the full measure of savings is achieved, new methods of control along with the new all-variable speed equipment configuration must be incorporated into the design. 

Thomas Hartman, P.E.,
The Hartman Company

In the past two months, I have presented articles intended to help owners and managers of large facilities understand and evaluate the benefits of "All-Variable Speed" chilled water plants and distribution systems. The application of All-Variable Speed cooling systems yields a substantial energy dividend because these systems vastly improve operating efficiency at part load conditions where cooling systems typically spend most of their operating hours. However, to ensure that the full measure of savings is achieved, new methods of control along with the new all-variable speed equipment configuration must be incorporated into the design. This month I will describe the new control concept called "Demand Based Control" that has been developed specifically for the optimization of all-variable speed systems. 

Problems with Current Control in All-Variable Speed Applications 

[an error occurred while processing this directive]

Variable speed operation of the centrifugal pumps, compressors and fans that are employed in all-variable speed chilled water plants and distribution systems have very different operating characteristics compared to constant speed equipment. The two differences that make variable-speed equipment unsuitable for conventional control techniques are:

  1. The potential for substantially higher operating efficiency at part load conditions.

  2. A reduction in head capacity of variable speed pumps, compressors, and fans that varies with the square of the speed reduction.

Control problems stemming from the higher operating efficiency at part load conditions can be demonstrated with the graph in Figure 1. This graph compares the "wire to water" efficiency for an optimally operated constant speed chiller plant and an optimally operated all-variable speed chiller plant of the same first cost at identical loading conditions. The term "wire to water" is used to denote that the efficiencies are based on the actual measured input power, which includes all mechanical and electrical losses such as motors, VFDs, and other losses.

In Figure 1, the cost of the all-variable speed plant is equalized with the constant speed plant by selecting equipment that is slightly less efficient for the all-variable speed plant to offset the costs associated with the VFD on each piece of equipment. The lower nominal efficiencies are characterized in the graph by the relative efficiency at the 100% load point. Here, the all-variable speed plant is less efficient than the constant speed plant because variable speed equipment of the same first cost as constant speed equipment has a reduced peak efficiency. However, it should also be noted that most chiller plants are built with some excess capacity to protect against failure of one or more components. When even a small amount of excess capacity is incorporated in the plant and distribution system designs, an all-variable speed plant will operate more efficiently than a constant speed plant even at full load conditions.

The bar graph in Figure 1 shows the expected operating hours at each loading for a typical chiller plant. Note that the majority of operating time is spent at loading ranges where the all-variable speed plant is significantly more efficient than the optimized constant speed plant.

Operating Efficiency Comparison

Figure 1: Comparison of Optimized Constant Speed and "All-Variable Speed" Chilled Water Plant Operating Efficiencies

Figure 1 thus illustrates the problem of employing traditional equipment sequencing controls for an all-variable speed system. Traditional controls invariably sequence equipment on and off line to keep each unit operating as near as possible to its full capacity. Note in Figure 1 that the efficiency curve for the constant speed system is nearly flat above 50% loading and shows reduced efficiency below 50%. When additional equipment is brought on line, the equipment will operate at lower loading which is usually less efficient. Thus equipment is typically switched on only when the on-line equipment cannot meet the current operating requirements. However, as shown in Figure 1, if this sequencing technique were to be applied to an all-variable speed plant, it would reduce its operating efficiency because full load operation is the least efficient operating point for an all-variable speed system.

The second control concern is the fact that variable speed operation of most typical equipment results in reduced head capacity of the unit as the speed is reduced. This concern is shown specifically for variable speed pumps in Figure 2.

Typical Chilled Water Pump Operation

Figure 2: Comparison of the System Curve for a Conventionally Controlled Chilled Water Distribution pump and the System Curve of Optimum Pump Efficiency

Figure 2 shows the expected system curve for a conventionally controlled chilled water distribution pump that maintains a specific end-of-the-line differential pressure setpoint. This curve (in red) is compared to the system curve that yields highest pump efficiency at all flows (this is the green line that is often referred to the pump's "natural curve"). Note that at part loads where the pumping system operates most of the time, the curve are far apart. This difference translates into substantial energy penalties for pumps (and fans) when operated with conventional independent PID pressure and temperature setpoint controls.

The Equal Marginal Performance Principle 

The performance problems described above when conventional controls are applied to all-variable speed systems mandate a new method of control to optimize the operation of these new systems. A network control principle called the "Equal Marginal Performance Principle" (or EMPP) points the way to a new, more effective control when all-variable speed systems are employed. The Equal Marginal Performance Principle states a very simple relationship between input power and optimized operation of multiple modulating components: the operation of a system composed of multiple modulating components is optimized when the marginal system output divided by marginal system input is the same for each component in the system.

To better visualize the Equal Marginal Performance Principle and how it is applied to a cooling system, consider the following figure.

Equal Marginal Performance Principle Diagram

Figure 3: Equal Marginal Performance Principle Diagram.

Imagine a simple chiller plant and distribution system as shown in Figure 1. Now, imagine that the knobs below each piece of equipment adjust the capacity of that element by changing its speed. Finally, assume instrumentation is provided to measure the system output and system input as shown by the meters on the right side of the Figure. How would you optimize this system at its current point of operation? The Equal Marginal Performance Principle reasons that one could do it by making small adjustments to the capacity of each element, one at a time, and noting the changes in system output and input. Then, system efficiency can be improved by 1) reducing the speed setting for those elements that show relatively small marginal system capacity change per unit input change, and 2) increasing the capacity setting for those that show larger changes in capacity per unit power change such that the total system output remains unchanged. This process of testing each element and resetting the system is then repeated until all elements have exactly the same marginal output per unit change in power input. At that point the system is fully optimized.

[an error occurred while processing this directive]Performing on-line adaptive optimization in this manner would be slow and imprecise at best, especially when one considers that the capacity requirements of the system are likely to be changing at all time. The true power of this simple principle as applied to variable speed equipment is that when all equipment is modulated with variable speed control, simple power relationships among the elements of a cooling and distribution system can be fixed over large operating ranges. Thus arises the concept of "Demand Based Control" in which components are directly coupled and coordinated with simple preset power relationships to meet the system load rather than operated as independent subsystems with more complex temperature, pressure, or on-line optimization controls. This new control also eliminates the need for extra equipment to decouple elements from one another. 

Demand Based Control 

Demand Based Control is a new method of network control developed specifically for all-variable speed systems which leverages the Equal Marginal Performance Principle and network control to provide a breakthrough in HVAC system control simplicity and efficiency. It uses preset operating relationships between the various elements of an all-variable speed system to very simply coordinate the operation for each piece of equipment to meet loading demands most efficiently at all times. The effect of these simple controls is more stable system operation and improved operating efficiency.

To gain insight into how Demand Based Control works, consider the type of large all-variable speed distribution system that was discussed in the last article. Such a system has variable speed primary pumps that are sized to pump chilled water through the chillers and the distribution mains, and variable speed booster pumps that are connected in series with the primary pumps and pump chilled water through the buildings or groups of loads each serves. Conventional control avoids direct connection configurations because the operation of one group of pumps will affect the other. Keeping each control loop independent of others is one reason decoupling loops are applied in most multiple stage pumping systems.

However, with Demand Based Control, simpler direct coupled configurations have many advantages. Control of each pump is accomplished based on direct power (or speed) relationships. Therefore, as the demand for flow increases, the speed of both pumps is increased together according to a preset algorithm that ensures optimal performance at all possible loading levels. In most cases the primary pump serves multiple booster pump circuits. In these instances the primary and booster pumps are operated to achieve the highest overall efficiency. When one booster pump requires additional flow to meet an increased demand, the relative demand in the other booster circuits is factored into the algorithm that coordinates the primary and booster pump response. With effective network control, such integrated control is possible, and employing Demand Based Control rules, it can be accomplished simply as well.

Another example of an effective application for Demand Based Control is as a replacement for PID control to operate the chilled water valves on the loads served by an all-variable speed distribution system. The rules for PID control add substantial first costs and energy costs to typical distribution systems. Consider that valve sizing rules dictate that between 25% and 50% of the total pressure drop between the supply and return headers should be absorbed by the control valve. This means that up to 50% of the total pumping power is expended just to meet the requirements of the PID controls. Furthermore, the rules require that the pressure differential at all loads be as uniform as possible and that nearly continuous repositioning of the valves be employed to achieve satisfactory control under all conditions. These rules add first costs by requiring oversized piping (and reverse return configurations in some applications), and they add maintenance costs due to the frequent valve repositioning.

All of these costs can be eliminated or enormously mitigated by replacing PID control with Demand Based Control and integrating this control into the operation of the entire building system. The independence component of PID valve sizing can be mitigated with network control that coordinates the operation of each valve with others. The controllability component of valve sizing is similarly mitigated with network control that coordinates distribution temperature and pressure with what is really required at all times. Repositioning frequency for each is reduced by the replacement of "cooling demand" rather than a immutable temperature as the controlling factor. The result is a less costly, more efficient and much more inherently reliable, stable and effective distribution system. Demand Based Control does increase the need for effective design and layout of both the mechanical system and the network controls. But the performance improvement is well worth the effort! 

Summary & Conclusion 

All-variable speed technologies offer a giant leap forward in system efficiency and system performance when applied to building cooling systems. But to capture these benefits necessitates the replacement of traditional controls with a new control strategy called Demand Based Control that has been developed specifically for all-variable speed systems. Although Demand Based Control is very different from conventional controls, its application requires only the same rule based application that engineers have learned to use when applying conventional controls. Learning and using Demand Based Control presents no serious problems for motivated designers, and offers many system performance rewards for their clients!

Additional information on technologies discussed in this article is available at Comments and questions may be addressed to Mr. Hartman at


1. Hartman, Tom, 2002, "All-Variable Speed Centrifugal Chiller Plants: Can We Make Our Plants More Efficient?,", March 
2. Hartman, Tom, 2002, "All-Variable Speed Chilled Water Distribution Systems: Optimizing Distribution Efficiency,", April 
3. Hartman, T, 2001, "Ultra-Efficient Cooling with Demand Based Control" HPAC Engineering December. 
4. Hartman, T.B. 1999, "Network Based Control of Fluid Distribution Systems," Renewable And Advanced Energy Systems For the 21st Century, Lahaina, Hawaii.

[an error occurred while processing this directive]
[Click Banner To Learn More]

[Home Page]  [The Automator]  [About]  [Subscribe ]  [Contact Us]


Want Ads

Our Sponsors